Yamaha Dx7 Kontakt

[Barton Ostby]

JoannchrWhen you have a moment, please screen shot your Garritan CFX settings windows and post them here as a few images. I think others might enjoy leveraging your efforts and experience. Thank you.

Additional comments regarding parameters : For best performance , I used Max engine = 1.5 GB and small pre- caching. Allocating more memory, causes occasional cracks. With these parameters it runs perfectly ; I use 44100 Khz & 128 Buffer size as I found the latency too good but less realistic, I think the 10ms including all components is the closest value to an acoustic. Also I don't add any additional reverb as I love the natural Abbey Road reverb which is quite stunning , you can adjust it with the Ambient Cursor in the advanced screen .

Also I use linear curve, as it is the best curve I found using a VPC1 for this particular VST. I have to use different curves when I use Noire Pianoteq or Ivory but with Garritan this works best. Obviously it depends on your keyboard .

n-player I forgot to mention that I also tested an approach which is quite popular in the plugin space , which is the bx_subfilter plugin. This is very popular because of the very high quality of the plugin to start with, because virtually it has no perceptible added latency , so it can used live and last but not least because it is free. It works remarkably well with Garritan CFX and you can eliminate unwanted bass frequencies with the high pass filter ( called " tight punch" ) while increasing or decreasing the level of sub you want to obtain with the 'low end' knob. There is a resonance adjustment as well ( best position I found was 'high' , 'extreme' being too much) . Really great with the classic default samples of Garritan. ( I use it within Logic Pro X and the aria player VST)

@Joannchr @dore_m @CyberGene

One weird idea that crossed my mind: if the pedalling of the Garritan CFX is that good, it may be interesting to replace the samples in the CFX folders with the samples of @dore_m pianos... Dore' pianos "Garritan edition" for all us Garritan piano owners.

Reverse enginnering of the folder structure is somehow simple because all the CFX samples are individual WAV files (although desguised with a ".data" suffix), and the different perspectives are stored in separate folders. I have not checked in depth but seems that each perspective contains 6 subfolders (pedal-up, pedal-down, two similar for una corda, and another two for releases); each subfolder contains 20 sub-subfolders (velocity layers, probably) and each layer contains 49 samples (so, seems that not all keys are sampled).

The naming convention is really strange and seemingly random, but once the structure is understood and copied, it is not impossible to create a python script that renames all @dore_m samples to the Garritan structure.

It wouldn't work well because I doubt I've recorded the exact notes that garritan uses. I'd have to use the DAW to resample from kontakt every wav file that is required. And if they recorded every note, then that would be a large undertaking.

If I recall correctly, the Garritan script is far more complicated than my SFZ scripts, yet I think the fundamental idea is the same with the pedaling. If you take a look at the thread I recently responded to, there is a decay curve shape and re-routing of the pedal CC. Garritan probably does the same, but possibly with a much more well tuned curve that matches their recording (?). Just guessing.

Well... I should have studied this better because I looked at the installer files (I had removed Garritan CFX from my laptop). It turns out that after reinstalling CFX, the installed files are very different ?... The ".data" files in the installation folder (that play when the file suffix is changed to ".wav") are processed during installation into ".audio" files that do not play at all.

A poster in a makemusic forum states:

"Garritan Samples are stored in a RAW data format. You can view them with a RAW sample editor (I.E. Audacity), but they ARE slightly encrypted. What happens is, samples are coded with various errors that get fixed in real time by the ARIA player if the library is properly registered. It can be simple things like fixing the proper root-pitch, or correcting the left and right phasing in a stereo sample, or even some irregular artifacts and such."

From the little that I know about Garritan, it doesn't look like a fun scripting job. Some of the concepts are interesting to me, I think they use "dummy" CC (continuous controllers) like variables, since real variables are not supported in SFZ. And then the data is organized in written arrays and not with equations (AFAIK). Scripting and testing seems like a nightmare. Maybe they developed some internal graphical tools to aid in design revisions.

Of course, if we put in Dore's samples, and record the output from Garritan, then we could in principle reverse engineer the changes done by Garritan, because we know what the correct output should be (Dore's original samples).

Den som mener seg rammet av urettmessig publisering, oppfordres til ta kontakt med redaksjonen. Pressens Faglige Utvalg (PFU) er et klageorgan oppnevnt av Norsk Presseforbund som behandler klager mot mediene i presseetiske sprsml.

3a8082e126